Burry's whole book, graded

I went back through every post from the last five weeks — twelve Trading Posts and deep-dives — and tracked every disclosed position. This is the full map, not just the picks we've already covered. Grades are mine, on the strength of each argument, not a prediction. Burry himself keeps repeating that shorting rarely pays and his book is "mostly long most of the time."

The lens behind every call

Four ideas drive everything:

1. The 2000 rerun. He sees a "clonal repeat" of the Jan–Mar 2000 internet top — NASDAQ-100, the SOX semis index, and mega-IPOs (SpaceX, Anthropic) tracking the 2000 pattern. His playbook mirrors his original Scion launch: buy hated non-AI value while shorting the mania's fulcrum points with cheap puts.

2. The valuation system. "IV15" intrinsic-value multiples, "Tragic Algebra" (penalizes stock-comp dilution), and Castle / Chapel / Stone moat tiers for AI durability.

3. The turnover signal. A stock that's traded 3–5× its share count since its peak has moved into "steadier hands" and may be near a bottom. His main timing tool for catching falling knives. (Yes — the same signal I red-teamed on June 18. Still holds.)

4. Machines as lead steer. Algos and foundation models now dominate trading — his evidence is Fibonacci levels "working" again — and thin options volume plus dealer-gamma dynamics set up violent air-pocket declines in crowded names like NVIDIA.

The full book (as disclosed)

Shorts / puts: NVDA (puts) · PLTR (covered half the stock short at ~$107, keeps puts) · ORCL (puts) · QQQ (puts) · SOXX (puts + shorted the ETF, his DRAM/memory-cycle vehicle) · TSLA (short) · CAT (short, opened Jun 30) · AMAT (short, opened Jun 30).

US / Western longs: LULU (flagship, avg ~$124.65) · ADBE · PYPL · FISV · VEEV · ZTS · SFM · FMCC & FNMA · MELI · HCA · BIRK · MSFT (via long-dated LEAPs — wants the common near $350).

Asia / China longs: Samsung Electronics (top-3 fund position, bought at tangible book) · Tencent · Meituan · JD · BABA · Haidilao · Haier. Watching: BYD (mid-HK$70s) and PDD's HK listing.

The Samsung tension: Samsung is simultaneously a top-3 long (cheap franchise at book value) AND part of his memory-glut short via SOXX. He's long the company, short the cycle's blow-off. Galaxy-brain or hedged — you decide.

🔥 Hottest theses

Short DRAM/memory via SOXX — A−. Built on executives declaring cyclicality "structurally" dead (his textbook contrarian sell signal), the HBM "3× wafer" glut mechanic, record margins, and the 1998–2003 analogue where DRAM fell 93% even with a price-fixing cartel. Risk: pure timing — early-2027 puts expire worthless if AI-inference hoarding persists.

Long Samsung — A− (best risk-defined long). Mechanical rule: buy at tangible book. Eight prior instances over 30 years, ~24.6% annualized since 1998. The entry rule removes the guesswork. Risk: it's already run off book, and it's the same memory cycle he's shorting elsewhere.

Long LULU — B+ (most detailed deep-dive). ~2.5× tangible book (lowest since 2009), under 10× earnings, ~$1.5B net cash, $1B buyback against a ~$12B cap, IV15 $155.81 (0.72×), Ross Stores 2000 doppelganger. Risk he names himself: capricious branded retail, and it hinges entirely on unproven new CEO Heidi O'Neill (arrives Sept 8).

Long FISV + payments basket — B+. The thesis we broke down on June 18 (and red-teamed): 99% core retention, record ~$8 EPS, ~1.4× IV15, FIUSD stablecoin optionality. CEO exit flagged as "thesis violation → re-evaluate, not sell."

Long MELI & HCA — B (quietly high-quality). MELI bought on the turnover signal (~246% of shares traded, ~39% off the high). HCA on his "buy at 10–12× earnings, period" rule — currently dinged by the ACA-subsidy lapse, which he calls noise that hurts competitors more.

🥶 Coldest theses

Short CAT — C− (coldest fresh trade). He admits he's never shorted it and it's "always done great on the long side." The entire basis is a chart extended above the 200-day moving average. No fundamental teardown, no valuation, no catalyst. This is the one I'd fade hardest.

Short CRWD — A as analysis, D as a trade. Devastating teardown: dead last of his 50 software names, 23× IV15, ~22% stock-comp with no buyback, ~4.5%/yr dilution. But he is NOT short it — only "considered" it — because at all-time highs it can double before it halves. His own "sand castle" warning in action.

Short PLTR — C+ (cooling). Still core (16× IV15), but covering half the stock short at $107 and leaning on puts is a tacit retreat on his most crowded short. No near-term catalyst.

China basket — B− (high conviction, structurally cold). Best argument: the bond market — tight CDS on Tencent/Alibaba/Meituan vs blown-out Oracle/CoreWeave — says these are far safer than equity prices imply. But he concedes the "uninvestable" label has real basis: VIE structures and CCP policy create genuine zero-tails, and delivery-war margin recovery is "a year or more away."

🇰🇷 Sidebar: actually buying Samsung

No normal US listing — you can't just click buy like Apple. Roughly best-to-easiest: (1) the Korea primary line 005930.KS via a broker with KRX access (Interactive Brokers is the usual retail route; trades in won, Korea hours) — this is the actual common Burry references. (2) The London GDR — settles USD/GBP, thinner liquidity. (3) US OTC unsponsored ADR SSNLF — one-click if your broker allows it, but grey-market thin with wide spreads; limit orders only. (4) EWY (Korea ETF) — liquid, but diluted with dozens of names and you can't run his tangible-book entry rule on it. To replicate the signal you'd track tangible book per share on the Korea line and buy at/below ~1.0×.